CN109636826A - Live pig weight method for measurement, server and computer readable storage medium - Google Patents
Live pig weight method for measurement, server and computer readable storage medium Download PDFInfo
- Publication number
- CN109636826A CN109636826A CN201811342794.4A CN201811342794A CN109636826A CN 109636826 A CN109636826 A CN 109636826A CN 201811342794 A CN201811342794 A CN 201811342794A CN 109636826 A CN109636826 A CN 109636826A
- Authority
- CN
- China
- Prior art keywords
- live pig
- image
- model
- weight
- neural network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000005259 measurement Methods 0.000 title claims abstract description 54
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 138
- 238000000638 solvent extraction Methods 0.000 claims abstract description 30
- 244000309464 bull Species 0.000 claims abstract description 27
- 238000003709 image segmentation Methods 0.000 claims abstract description 21
- 239000000284 extract Substances 0.000 claims abstract description 14
- 238000003062 neural network model Methods 0.000 claims description 90
- 230000002708 enhancing effect Effects 0.000 claims description 31
- 230000011218 segmentation Effects 0.000 claims description 27
- 238000012545 processing Methods 0.000 claims description 18
- 210000002569 neuron Anatomy 0.000 claims description 16
- 238000000605 extraction Methods 0.000 claims description 13
- 238000001514 detection method Methods 0.000 claims description 8
- 238000004364 calculation method Methods 0.000 claims 1
- 238000013473 artificial intelligence Methods 0.000 abstract description 4
- 238000013528 artificial neural network Methods 0.000 abstract description 3
- 230000008569 process Effects 0.000 description 12
- 230000000694 effects Effects 0.000 description 10
- 238000010586 diagram Methods 0.000 description 8
- 238000012795 verification Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 6
- 241000282887 Suidae Species 0.000 description 5
- 238000004590 computer program Methods 0.000 description 5
- 241000208340 Araliaceae Species 0.000 description 4
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 4
- 235000003140 Panax quinquefolius Nutrition 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 235000008434 ginseng Nutrition 0.000 description 4
- 230000009466 transformation Effects 0.000 description 4
- 241001465754 Metazoa Species 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 210000000746 body region Anatomy 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 238000009395 breeding Methods 0.000 description 1
- 230000001488 breeding effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000018109 developmental process Effects 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 235000016709 nutrition Nutrition 0.000 description 1
- 230000035764 nutrition Effects 0.000 description 1
- 235000015277 pork Nutrition 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
- G06T5/94—Dynamic range modification of images or parts thereof based on local image properties, e.g. for local contrast enhancement
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/13—Edge detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20036—Morphological image processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to artificial intelligence, a kind of live pig weight method for measurement, server and computer readable storage medium are disclosed, which comprises obtain the bull live pig depth image of acquisition;Image segmentation is carried out to extract live pig ontology image to live pig depth image according to pre-set image partitioning algorithm;Extract training characteristics from live pig ontology image, the training characteristics include it is following one or more: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust;Live pig weight appraising model is obtained according to the training characteristics and the training of corresponding live pig weight information;Live pig image to be measured is input to live pig weight appraising model and obtains the weight information of the live pig to be measured.The present invention is based on neural networks to train to obtain live pig weight measurement model, and intelligence can be realized using the model and measure live pig weight, measurement is convenient and efficient, can effectively liberate manpower.
Description
Technical field
The present invention relates to field of image processing more particularly to live pig weight method for measurement, server and computer-readable deposit
Storage media.
Background technique
China is always animal husbandry big country, and pork is also always the important foodstuffs on people's dining table.With economical at full speed
Development, the industrial structure of animal husbandry have also carried out very big adjustment, and pig raising also enters modernization, the pig raising of scale now
Field management becomes particularly important.
China's technology is quickly grown now, and especially artificial intelligence technology is taken its place in the front ranks of the world.By newest artificial intelligence
Technology will be conducive to the generalization process of China's artificial intelligence applied to animal husbandry.In feeding process of live pig, effectively produce
Key is exactly optimal growth rate and feed conversion rate to be maintained by continuously monitoring, and feed conversion rate then occupies pig raising cost
3/4 is even more.An important factor for weight is influence above-mentioned two index, and evaluation live pig nutrition, growing environment and health
The important evidence of condition.In the breeding process on pig farm, it is often necessary to weigh to live pig, feeding is checked by weight change
The process of supporting whether there is problem.The mode of traditional measurement live pig weight is more troublesome, and many live pigs are only when selling
The growth for being once unfavorable for monitoring live pig can just be claimed.And live pig is mismatched during weighing, and struggle will cause data
Error substantially increases the workload that live pig is gone too far.
Therefore, the workload measured during live pig weight how is reduced, realizes and conveniently measures live pig body
It is current urgent problem to be solved again.
Summary of the invention
In view of this, the present invention proposes a kind of live pig weight method for measurement, server and computer readable storage medium, only
By being taken pictures to live pig and live pig weight need to can be calculated according to live pig image, save human cost and measurement tasks
Amount.
Firstly, to achieve the above object, the present invention proposes a kind of server, and the server includes memory, processor,
The live pig weight measurement system that can be run on the processor, the live pig weight measurement system are stored on the memory
Following steps are realized when being executed by the processor:
Obtain the bull live pig depth image of acquisition;
According to pre-set image partitioning algorithm to the live pig depth image carry out image segmentation, by live pig ontology image from
It is split in the live pig depth image;
Extract training characteristics from the live pig ontology image, the training characteristics include one of following parameter or more
A: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust;
It is established according to the training characteristics and corresponding live pig weight information and trains to obtain a live pig weight appraising model;
And
Live pig image to be measured is input to the live pig weight appraising model and obtains the weight information of the live pig to be measured.
Optionally, when the live pig weight measurement system is executed by the processor, following steps are also realized:
Conspicuousness detection is carried out to the live pig depth image, and using the high depth areas of conspicuousness as region of interest
Domain;And
The masking-out value of the live pig depth image is generated according to the area-of-interest, and according to the masking-out value to described
Live pig depth image carries out enhancing processing, obtains enhancing treated live pig depth image.
Optionally, the step of image segmentation is carried out to the live pig depth image according to pre-set image partitioning algorithm packet
It includes:
Gray level image is converted by the live pig depth image, and threshold value point is carried out to the gray level image using Da-Jin algorithm
It cuts;
Contours extract is carried out to the gray level image after segmentation, and the profile of extraction is filled;And
Filled image is subjected to morphology opening operation, to remove the external interference component of live pig sheet.
Optionally, described to be established according to the training characteristics and corresponding live pig weight information and train to obtain a live pig body
The step of appraising model includes: again
The training characteristics and live pig weight information corresponding with the training characteristics are divided into training set and verifying collection, are built
A vertical neural network model is simultaneously trained the neural network model using the training set;
It is verified using the neural network model that the verifying collection completes training, it is accurate to obtain model estimation
Rate;
Judge whether the model estimation accuracy rate is less than preset threshold;
If the model estimation accuracy rate is not less than the preset threshold, the neural network that the training is completed
Model is as the live pig weight appraising model;And
If the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted,
And neural network model adjusted is trained again using the training set, until the mould that the verifying collection verifying obtains
Type estimates that accuracy rate is not less than the preset threshold;
Wherein, the parameter of the neural network model includes the neuron number of total number of plies and each layer.
In addition, to achieve the above object, the present invention also provides a kind of live pig weight method for measurement, it is applied to server, institute
The method of stating includes:
Obtain the bull live pig depth image of acquisition;
According to pre-set image partitioning algorithm to the live pig depth image carry out image segmentation, by live pig ontology image from
It is split in the live pig depth image;
Extract training characteristics from the live pig ontology image, the training characteristics include one of following parameter or more
A: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust;
It is established according to the training characteristics and corresponding live pig weight information and trains to obtain a live pig weight appraising model;
And
Live pig image to be measured is input to the live pig weight appraising model and obtains the weight information of the live pig to be measured.
Optionally, after the step of bull live pig depth image for obtaining acquisition further include:
Conspicuousness detection is carried out to the live pig depth image, and using the high depth areas of conspicuousness as region of interest
Domain;And
The masking-out value of the live pig depth image is generated according to the area-of-interest, and according to the masking-out value to described
Live pig depth image carries out enhancing processing, obtains enhancing treated live pig depth image.
Optionally, the step of image segmentation is carried out to the live pig depth image according to pre-set image partitioning algorithm packet
It includes:
Gray level image is converted by the live pig depth image, and threshold value point is carried out to the gray level image using Da-Jin algorithm
It cuts;
Contours extract is carried out to the gray level image after segmentation, and the profile of extraction is filled;And
Filled image is subjected to morphology opening operation, to remove the external interference component of live pig sheet.
Optionally, described to be established according to the training characteristics and corresponding live pig weight information and train to obtain a live pig body
The step of appraising model includes: again
The training characteristics and live pig weight information corresponding with the training characteristics are divided into training set and verifying collection, are built
A vertical neural network model is simultaneously trained the neural network model using the training set;
It is verified using the neural network model that the verifying collection completes training, it is accurate to obtain model estimation
Rate;
Judge whether the model estimation accuracy rate is less than preset threshold;
If the model estimation accuracy rate is not less than the preset threshold, the neural network that the training is completed
Model is as the live pig weight appraising model;And
If the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and
Neural network model adjusted is trained again using the training set, until the model that the verifying collection verifying obtains
Estimate that accuracy rate is not less than the preset threshold;
Wherein, the parameter of the neural network model includes the neuron number of total number of plies and each layer.
Optionally, the step of parameter of the adjustment neural network model includes:
Adjust the total number of plies and/or each layer of neuron number of the neural network model.
Further, to achieve the above object, the present invention also provides a kind of computer readable storage medium, the computers
Readable storage medium storing program for executing is stored with live pig weight measurement system, and the live pig weight measurement system can be held by least one processor
Row, so that at least one described processor is executed such as the step of above-mentioned live pig weight method for measurement.
Compared to the prior art, live pig weight method for measurement, server and computer-readable storage proposed by the invention
Medium shoots live pig using image capture device, extracts live pig image features and the characteristics of image according to extraction
Parameter constructs and trains to obtain live pig weight measurement model, and then measures model realization using trained live pig weight and intelligently measure
Live pig weight is surveyed, measurement is convenient and efficient, precision is high, can effectively liberate manpower.
Detailed description of the invention
Fig. 1 is the schematic diagram of the optional hardware structure of server one of the present invention;
Fig. 2 is the program module schematic diagram of live pig weight measurement system first embodiment of the present invention;
Fig. 3 is the program module schematic diagram of live pig weight measurement system second embodiment of the present invention;
Fig. 4 is the implementation process diagram of live pig weight method for measurement first embodiment of the present invention;
Fig. 5 is the implementation process diagram of live pig weight method for measurement second embodiment of the present invention.
Appended drawing reference:
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
In order to make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and embodiments, right
The present invention is further elaborated.It should be appreciated that described herein, specific examples are only used to explain the present invention, not
For limiting the present invention.Based on the embodiments of the present invention, those of ordinary skill in the art are not before making creative work
Every other embodiment obtained is put, shall fall within the protection scope of the present invention.
It should be noted that the description for being related to " first ", " second " etc. in the present invention is used for description purposes only, and cannot
It is interpreted as its relative importance of indication or suggestion or implicitly indicates the quantity of indicated technical characteristic.Define as a result, " the
One ", the feature of " second " can explicitly or implicitly include at least one of the features.In addition, the skill between each embodiment
Art scheme can be combined with each other, but must be based on can be realized by those of ordinary skill in the art, when technical solution
Will be understood that the combination of this technical solution is not present in conjunction with there is conflicting or cannot achieve when, also not the present invention claims
Protection scope within.
As shown in fig.1, being the schematic diagram of the optional hardware structure of application server 2 one of the present invention.
In the present embodiment, the application server 2 may include, but be not limited only to, and company can be in communication with each other by system bus
Connect memory 11, processor 12, network interface 13.It should be pointed out that Fig. 2 illustrates only the application clothes with component 11-13
Business device 2, it should be understood that being not required for implementing all components shown, the implementation that can be substituted is more or less
Component.
Wherein, the application server 2 can be rack-mount server, blade server, tower server or cabinet
Formula server etc. calculates equipment, which can be independent server, be also possible to composed by multiple servers
Server cluster.
The memory 11 include at least a type of readable storage medium storing program for executing, the readable storage medium storing program for executing include flash memory,
Hard disk, multimedia card, card-type memory (for example, SD or DX memory etc.), random access storage device (RAM), static random are visited
It asks memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), may be programmed read-only deposit
Reservoir (PROM), magnetic storage, disk, CD etc..In some embodiments, the memory 11 can be the application clothes
The internal storage unit of business device 2, such as the hard disk or memory of the application server 2.In further embodiments, the memory
11 are also possible to the plug-in type hard disk being equipped on the External memory equipment of the application server 2, such as the application server 2,
Intelligent memory card (Smart Media Card, SMC), secure digital (Secure Digital, SD) card, flash card (Flash
Card) etc..Certainly, the memory 11 can also both including the application server 2 internal storage unit and also including outside it
Portion stores equipment.In the present embodiment, the memory 11 is installed on the operating system of the application server 2 commonly used in storage
With types of applications software, such as the program code etc. of live pig weight measurement system 100.In addition, the memory 11 can also be used
In temporarily storing the Various types of data that has exported or will export.
The processor 12 can be in some embodiments central processing unit (Central Processing Unit,
CPU), controller, microcontroller, microprocessor or other data processing chips.The processor 12 is commonly used in answering described in control
With the overall operation of server 2.In the present embodiment, the processor 12 is for running the program generation stored in the memory 11
Code or processing data, such as run the live pig weight measurement system 100 etc..
The network interface 13 may include radio network interface or wired network interface, which is commonly used in
Communication connection is established between the application server 2 and other electronic equipments.
So far, oneself is through describing the hardware configuration and function of relevant device of the present invention in detail.In the following, above-mentioned introduction will be based on
It is proposed each embodiment of the invention.
Firstly, the present invention proposes a kind of live pig weight measurement system 100.
As shown in fig.2, being the Program modual graph of 100 first embodiment of live pig weight measurement system of the present invention.
In the present embodiment, the live pig weight measurement system 100 includes a series of calculating being stored on memory 11
The live pig weight of various embodiments of the present invention may be implemented when the computer program instructions are executed by processor 12 in machine program instruction
Amount measurement operation.In some embodiments, the specific operation realized based on the computer program instructions each section, live pig weight
Amount measurement system 100 can be divided into one or more modules.For example, in Fig. 2, live pig weight measurement system 100 can be with
It is divided into and obtains module 101, segmentation module 102, extraction module 103, model training module 104 and measurement module 105.Its
In:
The bull live pig depth image for obtaining module 101 and being used to obtain acquisition.
In one embodiment, it can shoot to obtain bull live pig depth image, the depth by depth image capture apparatus
Degree image picking-up apparatus can be the capture apparatus such as depth camera, mobile phone with depth image shooting function.The acquisition
Module 101 can be communicated to obtain the bull live pig depth image of acquisition with the depth image capture apparatus.The depth
Degree image picking-up apparatus, which shoots to obtain bull live pig depth image, also can store in a sample database, the acquisition module 101
Bull live pig depth image is obtained by being communicated with sample database.
The segmentation module 102 is used to carry out image point to the live pig depth image according to pre-set image partitioning algorithm
It cuts, live pig ontology image is split from the live pig depth image.
In one embodiment, the pre-set image partitioning algorithm can be gray threshold segmentation algorithm.The segmentation module
102 are split live pig ontology using gray threshold segmentation algorithm from the live pig depth image, specific partitioning scheme
It is as described below: firstly, converting gray level image for live pig depth image;Secondly, being carried out using Da-Jin algorithm to gray level image adaptive
Answer Threshold segmentation;Furthermore contours extract is carried out to the image after segmentation, and the profile of extraction is filled;Finally, to filling
Image afterwards carries out morphology opening operation, removes the external interference component of live pig sheet, and then obtain live pig ontology image.In this hair
In bright other embodiments, the segmentation module 102 can also take other to be able to achieve the algorithm of image segmentation to the life
Pig depth image carries out image segmentation, such as the partitioning algorithm based on edge, the partitioning algorithm based on region.
For extracting training characteristics from the live pig ontology image, the training characteristics include the extraction module 103
Following parameter one or more: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust.
In one embodiment, the parameter value of each training characteristics in live pig ontology image can be according to training characteristics institute
Distance computation between the number and pixel of the pixel accounted for obtains.
The model training module 104 is for establishing and training according to the training characteristics and corresponding live pig weight information
Obtain a live pig weight appraising model.
In one embodiment, the live pig weight appraising model can be one based on neural network model and a large amount of
Live pig training characteristics-weight sample data trains the model come.In those live pig depth images as training sample
There is known weight per first-born pig.The weight information can be an interval value, such as positive and negative the 2 of the weight actually measured are public
Jin.
For example, the neural network model includes input layer, multiple hidden layers and output layer.Input layer is for receiving
The sample data of the training characteristics of multiple live pigs and weight corresponding with the training characteristics.Each hidden layer includes corresponding multiple
Node (neuron), each node in each hidden layer are configured at least one to the adjacent lower in the model
The output of a node executes linear or nonlinear transformation.Wherein, the input of the node of upper layer hidden layer can be based under adjacent
The output of a node or several nodes in layer.Each hidden layer has corresponding weight, which is based on number of training
According to acquisition, when being trained to model, can be obtained by carrying out the pre-training of model using there is the learning process of supervision
To the initial weight of each hidden layer.It can be calculated by using back-propagation when the initial weight of each hidden layer is adjusted
Method carries out, and output layer is for receiving the output signal from the last layer hidden layer.
In one embodiment, the model training module 104 establishes the concrete mode of the live pig weight appraising model
Include:
D1. multiple training characteristics and corresponding live pig weight information are divided into training set and verifying collects, one mind of building
The neural network model is trained through network model and using the training set;
D2. it is verified using the neural network model that the verifying collection completes training, obtains model estimation
Accuracy rate;
D3. judge whether the model estimation accuracy rate is less than preset threshold, if model estimation accuracy rate is not less than
The preset threshold, the neural network model that the training is completed is as the live pig weight appraising model;
If D4. the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and
Neural network model adjusted is trained again using the training set, until the model estimation that verifying collection verifying obtains
Accuracy rate is not less than the preset threshold, wherein the parameter of the neural network model includes the neuron of total number of plies and each layer
Number.
In one embodiment, the model training module 104 can be by adjusting total layer of the neural network model
Several and/or each layers of neuron number is adjusted the neural network model to realize.
In one embodiment, the training set i.e. for being trained to neural network model, use by the verifying collection
It is verified in neural network model.Specifically, neural network model is trained to obtain one first with the training set
The data of the verifying collection are input to the mid-module and carry out Accuracy Verification, according to each verification result by mid-module
It can count to obtain model estimation accuracy rate, judge whether the model estimation accuracy rate is less than preset threshold.When the mould
Type estimates that accuracy rate is not less than the preset threshold, shows that current mid-module estimation effect is preferable, satisfies the use demand, can
Using by the mid-module as the live pig weight appraising model.When model estimation accuracy rate is less than the default threshold
Value shows that current mid-module estimation effect is bad, is improved, adjust the ginseng of the neural network model at this time
Number, and neural network model adjusted is trained again using the training set to obtain a new mid-module, then
The mid-module retrieved is verified using the verifying collection again to obtain a new model estimation accuracy rate.If should
New model estimation accuracy rate needs to repeat the above steps again until collecting by the verifying still less than the preset threshold
Obtained model estimation accuracy rate is not less than the preset threshold.
In one embodiment, the preset threshold can be set according to actual use demand.Such as described in setting
Preset threshold is 95%.
The measurement module 105 be used to for live pig image to be measured being input to the live pig weight appraising model obtain it is described to
Survey the weight information of live pig.
It in one embodiment, can be with after completing to be trained to live pig weight appraising model and obtain available model
It obtains and inputs live pig image to be measured, the output of model at this time is the weight information of the live pig to be measured.
As shown in fig.3, being the Program modual graph of 100 second embodiment of live pig weight measurement system of the present invention.This implementation
In example, the live pig weight measurement system 100 includes a series of computer program instructions being stored on memory 11, when this
When computer program instructions are executed by processor 12, the live pig weight metrology operation of various embodiments of the present invention may be implemented.One
In a little embodiments, based on the specific operation that the computer program instructions each section is realized, live pig weight measurement system 100 can
To be divided into one or more modules.For example, live pig weight measurement system 100 can be divided into acquisition module in Fig. 3
101, divide module 102, extraction module 103, model training module 104, measurement module 105 and enhancing module 106.Each journey
Sequence module 101-105 is identical as live pig weight 100 first embodiments of measurement system of the present invention, and increases enhancing mould on this basis
Block 106.Wherein:
The bull live pig depth image for obtaining module 101 and being used to obtain acquisition.
In one embodiment, it can shoot to obtain bull live pig depth image, the depth by depth image capture apparatus
Degree image picking-up apparatus can be the capture apparatus such as depth camera, mobile phone with depth image shooting function.The acquisition
Module 101 can be communicated to obtain the bull live pig depth image of acquisition with the depth image capture apparatus.The depth
Degree image picking-up apparatus, which shoots to obtain bull live pig depth image, also can store in a sample database, the acquisition module 101
Bull live pig depth image is obtained by being communicated with sample database.
The enhancing module 106 is used to carry out conspicuousness detection to the live pig depth image, by the high depth of conspicuousness
Region generates the masking-out value of the live pig depth image according to the area-of-interest as area-of-interest, and according to described
Masking-out value carries out enhancing processing to the live pig depth image.
In one embodiment, it when being shot by the depth image capture apparatus to live pig, can may also incite somebody to action
Environment locating for live pig or other garbages are filmed, at this time can be by the enhancing module 106 to the live pig
Depth image carries out image enhancement processing, to highlight to the live pig ontology in the live pig depth image, after raising
Continuous step carries out the accuracy of image segmentation.Specifically, the enhancing module 106 can be accomplished by the following way to live pig depth
Live pig ontology in degree image is highlighted: firstly, according to the depth value of each pixel of the live pig depth image
The live pig depth image is split, and then is partitioned into the different depth region of the live pig depth image;Secondly, by institute
It states live pig depth image and carries out conspicuousness detection, and using the high depth areas of conspicuousness as the live pig depth image
Area-of-interest, the area-of-interest are live pig body regions;Furthermore according to the region of interest of the live pig depth image
Domain carries out binary segmentation to the live pig depth image, wherein the masking-out value of the area-of-interest of the live pig depth image is 1,
The masking-out value of the live pig depth image regions of non-interest is 0;Finally, by the area-of-interest of the live pig depth image with
The adjacent boundary of the regions of non-interest of the live pig depth image is smoothed, and then obtains the illiteracy of regions of non-interest
Version value range is 0≤Mask < 1, and the region of Mask=0 without processing, or carries out Fuzzy processing, inhibits non-interested
Region;The region of Mask=1 carry out brightness enhancing, edge sharpening, contrast enhancing and color saturation enhancing one kind or
A variety of, the region of 0 < Mask < 1 carries out different degrees of enhancing processing, and Mask value is bigger, brightness enhancing, edge sharpening, right
Effect than degree enhancing and color saturation enhancing is stronger;Mask is smaller, and the degree of enhancing is weaker.
The segmentation module 102 is used to carry out image point to the live pig depth image according to pre-set image partitioning algorithm
It cuts, live pig ontology image is split from the live pig depth image.
In one embodiment, the pre-set image partitioning algorithm can be gray threshold segmentation algorithm.The segmentation module
102 are split live pig ontology using gray threshold segmentation algorithm from the live pig depth image, specific partitioning scheme
It is as described below: firstly, converting gray level image for live pig depth image;Secondly, being carried out using Da-Jin algorithm to gray level image adaptive
Answer Threshold segmentation;Furthermore contours extract is carried out to the image after segmentation, and the profile of extraction is filled;Finally, to filling
Image afterwards carries out morphology opening operation, removes the external interference component of live pig sheet, and then obtain live pig ontology image.In this hair
In bright other embodiments, the segmentation module 102 can also take other to be able to achieve the algorithm of image segmentation to the life
Pig depth image carries out image segmentation, such as the partitioning algorithm based on edge, the partitioning algorithm based on region.
For extracting training characteristics from the live pig ontology image, the training characteristics include the extraction module 103
Following parameter one or more: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust.
In one embodiment, the parameter value of each training characteristics in live pig ontology image can be according to training characteristics institute
Distance computation between the number and pixel of the pixel accounted for obtains.
The model training module 104 is for establishing and training according to the training characteristics and corresponding live pig weight information
Obtain a live pig weight appraising model.
In one embodiment, the live pig weight appraising model can be one based on neural network model and a large amount of
Live pig training characteristics-weight sample data trains the model come.In those live pig depth images as training sample
There is known weight per first-born pig.The weight information can be an interval value, such as positive and negative the 2 of the weight actually measured are public
Jin.
For example, the neural network model includes input layer, multiple hidden layers and output layer.Input layer is for receiving
The sample data of the training characteristics of multiple live pigs and weight corresponding with the training characteristics.Each hidden layer includes corresponding multiple
Node (neuron), each node in each hidden layer are configured at least one to the adjacent lower in the model
The output of a node executes linear or nonlinear transformation.Wherein, the input of the node of upper layer hidden layer can be based under adjacent
The output of a node or several nodes in layer.Each hidden layer has corresponding weight, which is based on number of training
According to acquisition, when being trained to model, can be obtained by carrying out the pre-training of model using there is the learning process of supervision
To the initial weight of each hidden layer.It can be calculated by using back-propagation when the initial weight of each hidden layer is adjusted
Method carries out, and output layer is for receiving the output signal from the last layer hidden layer.
In one embodiment, the model training module 104 establishes the concrete mode of the live pig weight appraising model
Include:
D1. multiple training characteristics and corresponding live pig weight information are divided into training set and verifying collects, one mind of building
The neural network model is trained through network model and using the training set;
D2. it is verified using the neural network model that the verifying collection completes training, obtains model estimation
Accuracy rate;
D3. judge whether the model estimation accuracy rate is less than preset threshold, if model estimation accuracy rate is not less than
The preset threshold, the neural network model that the training is completed is as the live pig weight appraising model;
If D4. the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and
Neural network model adjusted is trained again using the training set, until the model estimation that verifying collection verifying obtains
Accuracy rate is not less than the preset threshold, wherein the parameter of the neural network model includes the neuron of total number of plies and each layer
Number.
In one embodiment, the model training module 104 can be by adjusting total layer of the neural network model
Several and/or each layers of neuron number is adjusted the neural network model to realize.
In one embodiment, the training set i.e. for being trained to neural network model, use by the verifying collection
It is verified in neural network model.Specifically, neural network model is trained to obtain one first with the training set
The data of the verifying collection are input to the mid-module and carry out Accuracy Verification, according to each verification result by mid-module
It can count to obtain model estimation accuracy rate, judge whether the model estimation accuracy rate is less than preset threshold.When the mould
Type estimates that accuracy rate is not less than the preset threshold, shows that current mid-module estimation effect is preferable, satisfies the use demand, can
Using by the mid-module as the live pig weight appraising model.When model estimation accuracy rate is less than the default threshold
Value shows that current mid-module estimation effect is bad, is improved, adjust the ginseng of the neural network model at this time
Number, and neural network model adjusted is trained again using the training set to obtain a new mid-module, then
The mid-module retrieved is verified using the verifying collection again to obtain a new model estimation accuracy rate.If should
New model estimation accuracy rate needs to repeat the above steps again until collecting by the verifying still less than the preset threshold
Obtained model estimation accuracy rate is not less than the preset threshold.
In one embodiment, the preset threshold can be set according to actual use demand.Such as described in setting
Preset threshold is 95%.
The measurement module 105 be used to for live pig image to be measured being input to the live pig weight appraising model obtain it is described to
Survey the weight information of live pig.
It in one embodiment, can be with after completing to be trained to live pig weight appraising model and obtain available model
It obtains and inputs live pig image to be measured, the output of model at this time is the weight information of the live pig to be measured.
In addition, the present invention also proposes a kind of live pig weight method for measurement.
As shown in fig.4, being the implementation process diagram of live pig weight method for measurement first embodiment of the present invention.In this reality
It applies in example, the execution sequence of the step in flow chart shown in Fig. 4 can change according to different requirements, and certain steps can be with
It omits.
Step S400 obtains the bull live pig depth image of acquisition.
In one embodiment, it can shoot to obtain bull live pig depth image, the depth by depth image capture apparatus
Degree image picking-up apparatus can be the capture apparatus such as depth camera, mobile phone with depth image shooting function.Can with institute
Depth image capture apparatus is stated to be communicated to obtain the bull live pig depth image of acquisition.The depth image capture apparatus is clapped
Taking the photograph to obtain bull live pig depth image also can store in a sample database, and the acquisition module 101 with sample database by carrying out
Communication is to obtain bull live pig depth image.
Step S402 carries out image segmentation to the live pig depth image according to pre-set image partitioning algorithm, by live pig
Ontology image is split from the live pig depth image.
In one embodiment, the pre-set image partitioning algorithm can be gray threshold segmentation algorithm, utilize gray threshold
Partitioning algorithm splits live pig ontology from the live pig depth image, and specific partitioning scheme is as described below: firstly, will
Live pig depth image is converted into gray level image;Secondly, carrying out adaptive threshold fuzziness to gray level image using Da-Jin algorithm;Furthermore
Contours extract is carried out to the image after segmentation, and the profile of extraction is filled;Finally, carrying out form to filled image
Opening operation is learned, removes the external interference component of live pig sheet, and then obtain live pig ontology image.In other embodiments of the invention
In, other can also be taken to be able to achieve the algorithm of image segmentation to live pig depth image progress image segmentation, such as based on
The partitioning algorithm at edge, the partitioning algorithm based on region.
Step S404, extracts training characteristics from the live pig ontology image, and the training characteristics include following parameter
One or more: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust.
In one embodiment, the parameter value of each training characteristics in live pig ontology image can be according to training characteristics institute
Distance computation between the number and pixel of the pixel accounted for obtains.
Step S406 is established according to the training characteristics and corresponding live pig weight information and is trained to obtain a live pig weight
Appraising model.
In one embodiment, the live pig weight appraising model can be one based on neural network model and a large amount of
Live pig training characteristics-weight sample data trains the model come.In those live pig depth images as training sample
There is known weight per first-born pig.The weight information can be an interval value, such as positive and negative the 2 of the weight actually measured are public
Jin.
For example, the neural network model includes input layer, multiple hidden layers and output layer.Input layer is for receiving
The sample data of the training characteristics of multiple live pigs and weight corresponding with the training characteristics.Each hidden layer includes corresponding multiple
Node (neuron), each node in each hidden layer are configured at least one to the adjacent lower in the model
The output of a node executes linear or nonlinear transformation.Wherein, the input of the node of upper layer hidden layer can be based under adjacent
The output of a node or several nodes in layer.Each hidden layer has corresponding weight, which is based on number of training
According to acquisition, when being trained to model, can be obtained by carrying out the pre-training of model using there is the learning process of supervision
To the initial weight of each hidden layer.It can be calculated by using back-propagation when the initial weight of each hidden layer is adjusted
Method carries out, and output layer is for receiving the output signal from the last layer hidden layer.
In one embodiment, the specific steps for establishing the live pig weight appraising model may include:
D1. multiple training characteristics and corresponding live pig weight information are divided into training set and verifying collects, one mind of building
The neural network model is trained through network model and using the training set;
D2. it is verified using the neural network model that the verifying collection completes training, obtains model estimation
Accuracy rate;
D3. judge whether the model estimation accuracy rate is less than preset threshold, if model estimation accuracy rate is not less than
The preset threshold, the neural network model that the training is completed is as the live pig weight appraising model;
If D4. the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and
Neural network model adjusted is trained again using the training set, until the model estimation that verifying collection verifying obtains
Accuracy rate is not less than the preset threshold, wherein the parameter of the neural network model includes the neuron of total number of plies and each layer
Number.
It in one embodiment, can be by adjusting the total number of plies and/or each layer of nerve of the neural network model
First number is adjusted the neural network model to realize.
In one embodiment, the training set i.e. for being trained to neural network model, use by the verifying collection
It is verified in neural network model.Specifically, neural network model is trained to obtain one first with the training set
The data of the verifying collection are input to the mid-module and carry out Accuracy Verification, according to each verification result by mid-module
It can count to obtain model estimation accuracy rate, judge whether the model estimation accuracy rate is less than preset threshold.When the mould
Type estimates that accuracy rate is not less than the preset threshold, shows that current mid-module estimation effect is preferable, satisfies the use demand, can
Using by the mid-module as the live pig weight appraising model.When model estimation accuracy rate is less than the default threshold
Value shows that current mid-module estimation effect is bad, is improved, adjust the ginseng of the neural network model at this time
Number, and neural network model adjusted is trained again using the training set to obtain a new mid-module, then
The mid-module retrieved is verified using the verifying collection again to obtain a new model estimation accuracy rate.If should
New model estimation accuracy rate needs to repeat the above steps again until collecting by the verifying still less than the preset threshold
Obtained model estimation accuracy rate is not less than the preset threshold.
In one embodiment, the preset threshold can be set according to actual use demand.Such as described in setting
Preset threshold is 95%.
Live pig image to be measured is input to the live pig weight appraising model and obtains the body of the live pig to be measured by step S408
Weight information.
It in one embodiment, can be with after completing to be trained to live pig weight appraising model and obtain available model
It obtains and inputs live pig image to be measured, the output of model at this time is the weight information of the live pig to be measured.
As shown in fig.5, being the implementation process diagram of live pig weight method for measurement second embodiment of the present invention.In this reality
It applies in example, the execution sequence of the step in flow chart shown in fig. 5 can change according to different requirements, and certain steps can be with
It omits.
Step S400 obtains the bull live pig depth image of acquisition.
In one embodiment, it can shoot to obtain bull live pig depth image, the depth by depth image capture apparatus
Degree image picking-up apparatus can be the capture apparatus such as depth camera, mobile phone with depth image shooting function.Can with institute
Depth image capture apparatus is stated to be communicated to obtain the bull live pig depth image of acquisition.The depth image capture apparatus is clapped
Taking the photograph to obtain bull live pig depth image also can store in a sample database, and the acquisition module 101 with sample database by carrying out
Communication is to obtain bull live pig depth image.
Step S410 carries out conspicuousness detection to the live pig depth image, using the high depth areas of conspicuousness as sense
Interest region generates the masking-out value of the live pig depth image according to the area-of-interest, and according to the masking-out value to institute
It states live pig depth image and carries out enhancing processing.
In one embodiment, it when being shot by the depth image capture apparatus to live pig, can may also incite somebody to action
Environment locating for live pig or other garbages are filmed, and can carry out image enhancement to the live pig depth image at this time
Processing improves subsequent step and carries out image segmentation to highlight to the live pig ontology in the live pig depth image
Accuracy.Specifically, it can be accomplished by the following way and the live pig ontology in live pig depth image is highlighted: is first
First, the live pig depth image is split according to the depth value of each pixel of the live pig depth image, Jin Erfen
Cut out the different depth region of the live pig depth image;Secondly, the live pig depth image is carried out conspicuousness detection, and will
Area-of-interest of the high depth areas of conspicuousness as the live pig depth image, the area-of-interest is live pig
Body regions;Furthermore binary segmentation is carried out to the live pig depth image according to the area-of-interest of the live pig depth image,
Wherein the masking-out value of the area-of-interest of the live pig depth image is 1, the illiteracy of the live pig depth image regions of non-interest
Version value is 0;Finally, by the regions of non-interest of the area-of-interest of the live pig depth image and the live pig depth image
Adjacent boundary is smoothed, and then the masking-out value range for obtaining regions of non-interest is 0≤Mask < 1, the area of Mask=0
Domain without processing, or carries out Fuzzy processing, inhibits regions of non-interest;The region progress brightness enhancing of Mask=1,
Edge sharpening, contrast enhancing and color saturation enhancing it is one or more, the region of 0 < Mask < 1 carries out different journeys
The enhancing of degree is handled, and Mask value is bigger, brightness enhancing, edge sharpening, the effect of contrast enhancing and color saturation enhancing
It is stronger;Mask is smaller, and the degree of enhancing is weaker.
Step S402 carries out image segmentation to the live pig depth image according to pre-set image partitioning algorithm, by live pig
Ontology image is split from the live pig depth image.
In one embodiment, the pre-set image partitioning algorithm can be gray threshold segmentation algorithm, utilize gray threshold
Partitioning algorithm splits live pig ontology from the live pig depth image, and specific partitioning scheme is as described below: firstly, will
Live pig depth image is converted into gray level image;Secondly, carrying out adaptive threshold fuzziness to gray level image using Da-Jin algorithm;Furthermore
Contours extract is carried out to the image after segmentation, and the profile of extraction is filled;Finally, carrying out form to filled image
Opening operation is learned, removes the external interference component of live pig sheet, and then obtain live pig ontology image.In other embodiments of the invention
In, other can also be taken to be able to achieve the algorithm of image segmentation to live pig depth image progress image segmentation, such as based on
The partitioning algorithm at edge, the partitioning algorithm based on region.
Step S404, extracts training characteristics from the live pig ontology image, and the training characteristics include following parameter
One or more: live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust.
In one embodiment, the parameter value of each training characteristics in live pig ontology image can be according to training characteristics institute
Distance computation between the number and pixel of the pixel accounted for obtains.
Step S406 is established according to the training characteristics and corresponding live pig weight information and is trained to obtain a live pig weight
Appraising model.
In one embodiment, the live pig weight appraising model can be one based on neural network model and a large amount of
Live pig training characteristics-weight sample data trains the model come.In those live pig depth images as training sample
There is known weight per first-born pig.The weight information can be an interval value, such as positive and negative the 2 of the weight actually measured are public
Jin.
For example, the neural network model includes input layer, multiple hidden layers and output layer.Input layer is for receiving
The sample data of the training characteristics of multiple live pigs and weight corresponding with the training characteristics.Each hidden layer includes corresponding multiple
Node (neuron), each node in each hidden layer are configured at least one to the adjacent lower in the model
The output of a node executes linear or nonlinear transformation.Wherein, the input of the node of upper layer hidden layer can be based under adjacent
The output of a node or several nodes in layer.Each hidden layer has corresponding weight, which is based on number of training
According to acquisition, when being trained to model, can be obtained by carrying out the pre-training of model using there is the learning process of supervision
To the initial weight of each hidden layer.It can be calculated by using back-propagation when the initial weight of each hidden layer is adjusted
Method carries out, and output layer is for receiving the output signal from the last layer hidden layer.
In one embodiment, the specific steps for establishing the live pig weight appraising model may include:
D1. multiple training characteristics and corresponding live pig weight information are divided into training set and verifying collects, one mind of building
The neural network model is trained through network model and using the training set;
D2. it is verified using the neural network model that the verifying collection completes training, obtains model estimation
Accuracy rate;
D3. judge whether the model estimation accuracy rate is less than preset threshold, if model estimation accuracy rate is not less than
The preset threshold, the neural network model that the training is completed is as the live pig weight appraising model;
If D4. the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and
Neural network model adjusted is trained again using the training set, until the model estimation that verifying collection verifying obtains
Accuracy rate is not less than the preset threshold, wherein the parameter of the neural network model includes the neuron of total number of plies and each layer
Number.
It in one embodiment, can be by adjusting the total number of plies and/or each layer of nerve of the neural network model
First number is adjusted the neural network model to realize.
In one embodiment, the training set i.e. for being trained to neural network model, use by the verifying collection
It is verified in neural network model.Specifically, neural network model is trained to obtain one first with the training set
The data of the verifying collection are input to the mid-module and carry out Accuracy Verification, according to each verification result by mid-module
It can count to obtain model estimation accuracy rate, judge whether the model estimation accuracy rate is less than preset threshold.When the mould
Type estimates that accuracy rate is not less than the preset threshold, shows that current mid-module estimation effect is preferable, satisfies the use demand, can
Using by the mid-module as the live pig weight appraising model.When model estimation accuracy rate is less than the default threshold
Value shows that current mid-module estimation effect is bad, is improved, adjust the ginseng of the neural network model at this time
Number, and neural network model adjusted is trained again using the training set to obtain a new mid-module, then
The mid-module retrieved is verified using the verifying collection again to obtain a new model estimation accuracy rate.If should
New model estimation accuracy rate needs to repeat the above steps again until collecting by the verifying still less than the preset threshold
Obtained model estimation accuracy rate is not less than the preset threshold.
In one embodiment, the preset threshold can be set according to actual use demand.Such as described in setting
Preset threshold is 95%.
Live pig image to be measured is input to the live pig weight appraising model and obtains the body of the live pig to be measured by step S408
Weight information.
It in one embodiment, can be with after completing to be trained to live pig weight appraising model and obtain available model
It obtains and inputs live pig image to be measured, the output of model at this time is the weight information of the live pig to be measured.
The serial number of the above embodiments of the invention is only for description, does not represent the advantages or disadvantages of the embodiments.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, technical solution of the present invention substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in a storage medium
In (such as ROM/RAM, magnetic disk, CD), including some instructions are used so that a terminal device (can be mobile phone, computer, clothes
Business device, air conditioner or the network equipment etc.) execute method described in each embodiment of the present invention.
The above is only a preferred embodiment of the present invention, is not intended to limit the scope of the invention, all to utilize this hair
Equivalent structure or equivalent flow shift made by bright specification and accompanying drawing content is applied directly or indirectly in other relevant skills
Art field, is included within the scope of the present invention.
Claims (10)
1. a kind of live pig weight method for measurement is applied to server, which is characterized in that the described method includes:
Obtain the bull live pig depth image of acquisition;
Image segmentation is carried out to the live pig depth image according to pre-set image partitioning algorithm, by live pig ontology image from described
It is split in live pig depth image;
Training characteristics are extracted from the live pig ontology image, the training characteristics include one or more of following parameter:
Live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust;
It is established according to the training characteristics and corresponding live pig weight information and trains to obtain a live pig weight appraising model;And
Live pig image to be measured is input to the live pig weight appraising model and obtains the weight information of the live pig to be measured.
2. live pig weight method for measurement as described in claim 1, which is characterized in that the bull live pig depth for obtaining acquisition
After the step of image further include:
Conspicuousness detection is carried out to the live pig depth image, and using the high depth areas of conspicuousness as area-of-interest;And
The masking-out value of the live pig depth image is generated according to the area-of-interest, and according to the masking-out value to the live pig
Depth image carries out enhancing processing, obtains enhancing treated live pig depth image.
3. live pig weight method for measurement as claimed in claim 1 or 2, which is characterized in that described to divide calculation according to pre-set image
Method to the live pig depth image carry out image segmentation the step of include:
Gray level image is converted by the live pig depth image, and Threshold segmentation is carried out to the gray level image using Da-Jin algorithm;
Contours extract is carried out to the gray level image after segmentation, and the profile of extraction is filled;And
Filled image is subjected to morphology opening operation, to remove the external interference component of live pig sheet.
4. live pig weight method for measurement as claimed in claim 1 or 2, which is characterized in that it is described according to the training characteristics and
Corresponding live pig weight information is established and trains the step of obtaining a live pig weight appraising model
The training characteristics and live pig weight information corresponding with the training characteristics are divided into training set and verifying collection, establish one
Neural network model is simultaneously trained the neural network model using the training set;
It is verified using the neural network model that the verifying collection completes training, obtains model estimation accuracy rate;
Judge whether the model estimation accuracy rate is less than preset threshold;
If the model estimation accuracy rate is not less than the preset threshold, the neural network model that the training is completed
As the live pig weight appraising model;And
If the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model is adjusted, and utilize
The training set is again trained neural network model adjusted, until the model estimation that the verifying collection verifying obtains
Accuracy rate is not less than the preset threshold;
Wherein, the parameter of the neural network model includes the neuron number of total number of plies and each layer.
5. live pig weight method for measurement as claimed in claim 4, which is characterized in that the adjustment neural network model
The step of parameter includes:
Adjust the total number of plies and/or each layer of neuron number of the neural network model.
6. a kind of server, which is characterized in that the server includes memory, processor, and being stored on the memory can
The live pig weight measurement system run on the processor, it is real when the live pig weight measurement system is executed by the processor
Existing following steps:
Obtain the bull live pig depth image of acquisition;
Image segmentation is carried out to the live pig depth image according to pre-set image partitioning algorithm, by live pig ontology image from described
It is split in live pig depth image;
Training characteristics are extracted from the live pig ontology image, the training characteristics include one or more of following parameter:
Live pig Ontology project area, live pig body are long, live pig body is high, live pig body is wide, live pig bust;
It is established according to the training characteristics and corresponding live pig weight information and trains to obtain a live pig weight appraising model;And
Live pig image to be measured is input to the live pig weight appraising model and obtains the weight information of the live pig to be measured.
7. server as claimed in claim 6, which is characterized in that the live pig weight measurement system is executed by the processor
When, also realize following steps:
Conspicuousness detection is carried out to the live pig depth image, and using the high depth areas of conspicuousness as area-of-interest;And
The masking-out value of the live pig depth image is generated according to the area-of-interest, and according to the masking-out value to the live pig
Depth image carries out enhancing processing, obtains enhancing treated live pig depth image.
8. server as claimed in claims 6 or 7, which is characterized in that it is described according to pre-set image partitioning algorithm to the life
Pig depth image carry out image segmentation the step of include:
Gray level image is converted by the live pig depth image, and Threshold segmentation is carried out to the gray level image using Da-Jin algorithm;
Contours extract is carried out to the gray level image after segmentation, and the profile of extraction is filled;And
Filled image is subjected to morphology opening operation, to remove the external interference component of live pig sheet.
9. server as claimed in claims 6 or 7, which is characterized in that described according to the training characteristics and corresponding live pig
Weight information is established and trains the step of obtaining a live pig weight appraising model
The training characteristics and live pig weight information corresponding with the training characteristics are divided into training set and verifying collection, establish one
Neural network model is simultaneously trained the neural network model using the training set;
It is verified using the neural network model that the verifying collection completes training, obtains model estimation accuracy rate;
Judge whether the model estimation accuracy rate is less than preset threshold;
If the model estimation accuracy rate is not less than the preset threshold, the neural network model that the training is completed
As the live pig weight appraising model;And
If the model estimation accuracy rate is less than the preset threshold, the parameter of the neural network model, and benefit are adjusted
Neural network model adjusted is trained again with the training set, until the model that the verifying collection verifying obtains is estimated
It calculates accuracy rate and is not less than the preset threshold;
Wherein, the parameter of the neural network model includes the neuron number of total number of plies and each layer.
10. a kind of computer readable storage medium, the computer-readable recording medium storage has live pig weight measurement system, institute
Stating live pig weight measurement system can be executed by least one processor, so that at least one described processor executes such as claim
Described in any one of 1-5 the step of live pig weight method for measurement.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811342794.4A CN109636826A (en) | 2018-11-13 | 2018-11-13 | Live pig weight method for measurement, server and computer readable storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811342794.4A CN109636826A (en) | 2018-11-13 | 2018-11-13 | Live pig weight method for measurement, server and computer readable storage medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109636826A true CN109636826A (en) | 2019-04-16 |
Family
ID=66067907
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811342794.4A Pending CN109636826A (en) | 2018-11-13 | 2018-11-13 | Live pig weight method for measurement, server and computer readable storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109636826A (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110378241A (en) * | 2019-06-25 | 2019-10-25 | 北京百度网讯科技有限公司 | Crop growthing state monitoring method, device, computer equipment and storage medium |
CN110426112A (en) * | 2019-07-04 | 2019-11-08 | 平安科技(深圳)有限公司 | Live pig weight measuring method and device |
CN110612921A (en) * | 2019-09-25 | 2019-12-27 | 农芯科技(广州)有限责任公司 | Monitoring system and method for positioning gilts |
CN110672189A (en) * | 2019-09-27 | 2020-01-10 | 北京海益同展信息科技有限公司 | Weight estimation method, device, system and storage medium |
CN111507432A (en) * | 2020-07-01 | 2020-08-07 | 四川智迅车联科技有限公司 | Intelligent weighing method and system for agricultural insurance claims, electronic equipment and storage medium |
CN111724355A (en) * | 2020-06-01 | 2020-09-29 | 厦门大学 | Image measuring method for abalone body type parameters |
CN112330677A (en) * | 2021-01-05 | 2021-02-05 | 四川智迅车联科技有限公司 | High-precision weighing method and system based on image, electronic equipment and storage medium |
CN112801118A (en) * | 2021-02-26 | 2021-05-14 | 潘志乐 | Pork pig slaughtering benefit evaluation system and method based on artificial intelligence and big data |
CN113096178A (en) * | 2021-04-25 | 2021-07-09 | 中国农业大学 | Pig weight estimation method, device, equipment and storage medium |
CN113532616A (en) * | 2020-04-15 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Weight estimation method, device and system based on computer vision |
CN113678786A (en) * | 2021-08-19 | 2021-11-23 | 陆荣清 | Ecological breeding method for improving disease resistance of live pigs |
CN116416260A (en) * | 2023-05-19 | 2023-07-11 | 四川智迅车联科技有限公司 | Weighing precision optimization method and system based on image processing |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101144705A (en) * | 2007-07-25 | 2008-03-19 | 中国农业大学 | Method for monitoring pig growth using binocular vision technology |
-
2018
- 2018-11-13 CN CN201811342794.4A patent/CN109636826A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101144705A (en) * | 2007-07-25 | 2008-03-19 | 中国农业大学 | Method for monitoring pig growth using binocular vision technology |
Non-Patent Citations (4)
Title |
---|
刘同海 等: "基于RBF神经网络的种猪体重预测", 《农业机械学报》, vol. 44, no. 08, pages 245 - 249 * |
张凯 等: "基于计算机视觉技术育肥猪体重分析研究", 《农机化研究》, no. 05, pages 32 - 36 * |
李卓 等: "基于深度图像的猪体尺检测***", 《农业机械学报》, vol. 47, no. 03, pages 311 - 318 * |
王琳 等: "基于深度图像和BP 神经网络的肉鸡体质量估测模型", 《农业工程学报》, vol. 33, no. 13, pages 199 - 205 * |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110378241A (en) * | 2019-06-25 | 2019-10-25 | 北京百度网讯科技有限公司 | Crop growthing state monitoring method, device, computer equipment and storage medium |
CN110378241B (en) * | 2019-06-25 | 2022-04-29 | 北京百度网讯科技有限公司 | Crop growth state monitoring method and device, computer equipment and storage medium |
CN110426112A (en) * | 2019-07-04 | 2019-11-08 | 平安科技(深圳)有限公司 | Live pig weight measuring method and device |
CN110612921A (en) * | 2019-09-25 | 2019-12-27 | 农芯科技(广州)有限责任公司 | Monitoring system and method for positioning gilts |
CN110672189A (en) * | 2019-09-27 | 2020-01-10 | 北京海益同展信息科技有限公司 | Weight estimation method, device, system and storage medium |
CN113532616A (en) * | 2020-04-15 | 2021-10-22 | 阿里巴巴集团控股有限公司 | Weight estimation method, device and system based on computer vision |
CN111724355A (en) * | 2020-06-01 | 2020-09-29 | 厦门大学 | Image measuring method for abalone body type parameters |
CN111724355B (en) * | 2020-06-01 | 2022-06-14 | 厦门大学 | Image measuring method for abalone body type parameters |
CN111507432A (en) * | 2020-07-01 | 2020-08-07 | 四川智迅车联科技有限公司 | Intelligent weighing method and system for agricultural insurance claims, electronic equipment and storage medium |
CN112330677A (en) * | 2021-01-05 | 2021-02-05 | 四川智迅车联科技有限公司 | High-precision weighing method and system based on image, electronic equipment and storage medium |
CN112801118A (en) * | 2021-02-26 | 2021-05-14 | 潘志乐 | Pork pig slaughtering benefit evaluation system and method based on artificial intelligence and big data |
CN112801118B (en) * | 2021-02-26 | 2024-02-02 | 潘志乐 | Pork pig marketing benefit evaluation system and method based on artificial intelligence and big data |
CN113096178A (en) * | 2021-04-25 | 2021-07-09 | 中国农业大学 | Pig weight estimation method, device, equipment and storage medium |
CN113678786A (en) * | 2021-08-19 | 2021-11-23 | 陆荣清 | Ecological breeding method for improving disease resistance of live pigs |
CN116416260A (en) * | 2023-05-19 | 2023-07-11 | 四川智迅车联科技有限公司 | Weighing precision optimization method and system based on image processing |
CN116416260B (en) * | 2023-05-19 | 2024-01-26 | 四川智迅车联科技有限公司 | Weighing precision optimization method and system based on image processing |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109636826A (en) | Live pig weight method for measurement, server and computer readable storage medium | |
CN104850836B (en) | Insect automatic distinguishing method for image based on depth convolutional neural networks | |
CN107667903B (en) | Livestock breeding living body weight monitoring method based on Internet of things | |
US20230281265A1 (en) | Method for estimating body size and weight of pig based on deep learning | |
CN109272016A (en) | Object detection method, device, terminal device and computer readable storage medium | |
CN109657564A (en) | A kind of personnel detection method, device, storage medium and terminal device on duty | |
CN108805137A (en) | Extracting method, device, computer equipment and the storage medium of livestock feature vector | |
CN110083723B (en) | Small panda individual identification method, equipment and computer readable storage medium | |
CN110135231A (en) | Animal face recognition methods, device, computer equipment and storage medium | |
CN106529006A (en) | Depth image-based broiler growth model fitting method and apparatus | |
CN110991222B (en) | Object state monitoring and sow oestrus monitoring method, device and system | |
CN108074236A (en) | Irrigating plant based reminding method, device, equipment and storage medium | |
CN112182269B (en) | Training of image classification model, image classification method, device, equipment and medium | |
CN104007733B (en) | It is a kind of that the system and method being monitored is produced to intensive agriculture | |
CN110175503A (en) | Length acquisition methods, device, settlement of insurance claim system, medium and electronic equipment | |
CN113420709A (en) | Cattle face feature extraction model training method and system and cattle insurance method and system | |
CN113421192A (en) | Training method of object statistical model, and statistical method and device of target object | |
CN109460792A (en) | A kind of artificial intelligence model training method and device based on image recognition | |
CN116486296A (en) | Target detection method, device and computer readable storage medium | |
CN108182444A (en) | The method and device of video quality diagnosis based on scene classification | |
CN112597995B (en) | License plate detection model training method, device, equipment and medium | |
CN109359675A (en) | Image processing method and equipment | |
CN113838076A (en) | Method and device for labeling object contour in target image and storage medium | |
CN116630828A (en) | Unmanned aerial vehicle remote sensing information acquisition system and method based on terrain environment adaptation | |
CN110163253A (en) | Fish floating head degree detecting method and system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |